Slashdot Mirror
Terabyte Hard Drive Put To the Test
EconolineCrush writes "As a technical milestone, Hitachi's Deskstar 7K1000 hard drive is undeniably impressive. The drive is the first to pack a trillion bytes into a standard 3.5" form factor, and while some may argue the merits of tebi versus tera, that's still an astounding accomplishment. Hitachi also outfitted the drive with 32MB of cache—double what you get with standard desktop drives—making this latest Deskstar a leader in both cache size and total capacity. That looks like a great formula for success on paper, but how does it pan out in the real world? The Tech Report has tested the 7K1000's performance, noise levels, and power consumption against 18 other drives to find out, with surprising results."
FTFA: "Gigabyte drives were only "missing" 24 bytes, and that was easy to swallow."
i think they meant 24 megabytes, which is easy to scoff at now, but wasn't when the first gigabyte drives dropped.
Beware the Jubjub bird, and shun the frumious Bandersnatch.
This marketing BS always pisses me off. For years and years and years we've used 1024 in the computer world, since it's a power of 2, and computers deal with powers of 2. A 931GB drive is NOT a 1TB drive. And we don't need new stupid labels like tebi, we just need storage manufacturers to stop being retards.
RAID 1+0 is the way to go for redundancy. Unless you're unlucky enough to lose both drives in one of the pairs making up the array, you can survive more than one drive failing.
It's also the way to go for speed - your controller doesn't have to calculate the parity bits for every write operation (yes I know the parity sum is simple - that doesn't stop it from adding a bottleneck).
RAID5 is most useful where:
1. You desperately need the space.
AND
2. You can't afford the drives (or, for that matter, power/larger RAID controller) required to acheive the same space in RAID 1+0.
Is there any point to these "huge" caches? My Linux system uses a few hundred MB's as disk cache so I don't really expext another few MB's on the disk to make any noticable difference (and, if I recall it correctly, when disks with 8 MB caches were new they did not really gave any performance advantage compared to models with only 2 MB of cache).
Real life is overrated.
I'm not that convinced by the testing methods here. The boot and load times page shows 20 seconds difference between the slowest and fastest drives which they barely comment on, and yet the drive with the slowest boot time is among the quickest when loading Far Cry and Doom 3? Something is not right there.
And if they're really timing level loads with a stopwatch, why on earth are they quoting 2 decimal places (and besides, the variability in reaction time is accounting for most of the supposed differences in any case). Half of their tests don't appear to tell anybody anything significant, and the most worthwhile page in there is the conclusion. Pretty graphics though.
Nothing new, then. At this point 1 TB may sound like "that much data", but then so did a 40 MB drive waaay back. Heck, at one point 1.4 MB meant a hard drive the size of a large washing machine. Nowadays that's called a floppy and already outdated.
What I'm getting at is that it's sorta like "Moore's law" for hard drives. (And occasionally Murphy's law too;) What's "whoa, I'd hate to lose that much data" at one point, is just adequate in a couple of years, and not even enough for your system files and/or swap file in 20 years.
A polar bear is a cartesian bear after a coordinate transform.
Look, I hate marketing dishonesty as much as the next guy, but borrowing the SI prefixes honestly does nothing but add confusion. Hard drives are easy, because one can safely assume that the marketing 'tards went with whatever number was bigger. But what about my phone's data plan? Aside from the whole kB vs kb thing, how do I know which definition of "kilo" my provider has gone with? Do they consider themselves with the "computer industry" or with the rest of the world? And (this is the best question), will the not-very-well-paid support grunt even know the difference?
Would you like it if you agreed to sell a dozen POS systems to a bakery, only to be told after the contract, "Sorry sir. This is the baking industry. You agreed to give us thirteen systems." Or if you got a $30 bill from your ISP with the explanation, "This is the computer industry. Though our adverts say this plan is $30 a month, that's hex. In base-ten dollars, you owe us $48."
You hate marketing people skewing reality. Good. It is only through fighting ambiguity that they can be stopped from getting away with this.
Do you know the difference between a pipe and a tube? If you get into any business involving either, I hope you don't repurpose the words everyone else has settled upon.
It's that extra bit of humility that really makes your post shine."Live as if you'll die tomorrow." Ridiculous. You could die later today.
Yeah, I've got to agree with you. I think that is one of the worst advices I have read on slashdot... A hard disk died on me a month after the S.M.A.R.T. thing started to annoy... it was on a laptop. Fortunately, I bought a bigger driver and passed all the information before the defective drive went dead.
While I agree that the S.M.A.R.T. heuristics might be a bit sensitive but if you consider what is at stake (yeah... your valuable pr0n collection), then I guess its better safe than sorry.
And, comparing it to the ink cartdriges? I am sure *your life* (or work...) does not depend on printing or not that pr0n picture...
Ubuntu is an African word meaning 'I can't configure Debian'
If you notice, everything listed in the parent is in powers of 10 bits (or Hz) except for disc capacities. Like everyone else said, this is because disc manufacturers want to confuse you. When talking about m/g/k bits the convention is to use powers of 10, and when talking about bytes it is to use powers of 2. Hence, as the parent said, powers of 2 are used for file sizes and RAM sizes... because those are usually in bytes.
But instead of going with whatever number that fits their specific field, they all went with 1000. Really, that IT people refuse to do the same makes us look utterly retarded.
Not that it matters anyway. With 8 bits on the byte, we're doomed before we even start. There is no hope in sight until we just ditch this shit, get a clue from the network people, and start counting bits in multiples of 1000.
I lost my sig.
No. I don't think you are. Or you are, but for a different reason.
Even the NSA very very probably can not recover any useful information from a disk overwritten the way I wrote. They have lots of money and expertise, but the laws of physics apply to them too.
But they could get at the information on your computer by other means that you'd be unlikely to detect, if they really wanted to. For example, if the information is from the net and you don't encrypt everything, they could easily wiretap your broadband. Getting a hardware-keylogger into your keyboard would be possible too, aswell as dozens of other tricks.
I've always thought you have a slightly better chance of getting valid data off of a drive if you never actually power it down when it's failing. This is anecdotal from a power outage causing many old hard drives in a building to give up, with their computers normally having uptime measured in months or even years.
Of course, to recover data like this you would need another computer accessible via the network, rather than installing a replacement in the desktop itself. Read any possible data off it while you still can, without putting it through the stress of powerdown/powerup.
"The universe seems neither benign nor hostile, merely indifferent." --Carl Sagan
The major disadvantage with hardware raid cards, though, is proprietary/undocumented on-disk formats. That sucks.
It died a horrible death only three years later, just outside of warranty. Despite a class action lawsuit against IBM (in the US, not Canada) I couldn't get it replaced. There was apparently a fix for it, simply by downloading a program, but really, who looks for updates to their hard drives?
IBM further went into my bad books, after it simply sold off the business to Hitachi instead of fixing their mess. It really left a sour taste in my mouth for IBM ...
No, when talking about RAM, where a MB is 1024 KB where a KB is 1024 bytes, you're talking about stuff connected to a memory controller that addresses this in a certain number of two, so that a 32 bit controller can address 4,294,967,296 bytes or 4 GiB. A disk controller works in a different way, and a disk is addressed in a different way. The only reason for demanding the same kind of numbering from a disk is when you need to know how much RAM a file will consume when you load it. Which is why a file's size may be denoted in KiB.
It really isn't confusing at all. I suspect the outrage at hard drive capacities is really caused by the high frequency of autism in the geek community.
So, what, you'd rather have had zero indications of hard disk failure then only one?
I've had four drives fail on me before (all of them Maxtor), SMART predicted one of them a month in advance by which time I'd backed up the whole thing. Maybe it missed the other three but even if it only catches a few errors, that's still a hell of a lot better than none isn't it?
Spelling mistakes, grammatical errors, and stupid comments are intentional.
Except they'd have more parts, more complexity, and the larger components would need to be made to even finer tolerences since they need to remain well aligned over a much larger area (and they'd need to be stronger if you wanted to keep the same sort of RPM). They'd be much more expensive, and you'd probably still have to drop the density per platter a lot to keep it within the realm of sanity, not least because of things like thermal expansion having a much larger effect.
File next to the disk with multiple drive head assemblies; possible, but just not worth it when you could just fit more, smaller, cheaper, independent disks in the same space.
TOO states " As the first hard drive to reach the terabyte mark, Hitachi's Deskstar 7K1000 will be remembered, too. Squeezing a trillion bytes into a 3.5" hard drive form factor is a monumental engineering achievement"
I doubt that anyone will remember this in a year. Quick; what was the model and manufacturer of the first drive to pass 500GB, or 1GB. Both were monumental engineering achievements in their time. These milestones will not be remembered because they are all evolutionary; a 10-30% jump in capacity. When we see 10x capacity increases in one generation, THAT name might be remembered.
That said.. good job Hitachi, but we all know that WD and Seagate will be out with their versions in a month or so.
Article X: The powers not delegated... by the Constitution...are reserved...to the people
Hmm... lets see... The IDE interface has been around since around 1984 so that's around 24 years.. Quite impressive... SCSI came at 1981 so that's around 26 years... Depending on the adapter you might even be able to access one of those old SASI disks that came around 1979, so now we are up to about 28-29 years of old hardware that you can access with currently available controllers... And you could even read punchcards with current hardware.. Just get a scanner :)
:)
:)
The issue is always if you have some possibility to read the media in the future, and it's always a hard thing if you have something that requires some extra reader with moving parts that can fail even if it's not being used due to corrosion and such... And i do think that a plain controller-card without any moving parts can be a bit easier to store, and if just using lots and lots of disk you can just migrate the data as time go by and the disks becomes cheaper.. Alot more fuss if you would want to migrate any of those LTO tapes to some new tape since that would require someone to fetch the tape, put it into a reader, read the data back, verify the data and then continue on with the next tape.. Just look at the past.. The amount of storage the disks have has exploded.. I remember when i got my first 'big' drive of a whole whopping 20MB and now around 20 years later i could fit 250 copies of that in my RAM..
So the problem is not really 'how do we store all the data' but more 'How do we migrate the data to new storage in the future?'
1. Man paints on cavewalls - still visible ~30000 years
2. Man carves on stone-platters - still visible after ~20000 years
3. Man writes on paper/papyrus etc - still readable after ~7500 years
4. Man invents computers - All unmaintained data older than 40 years is lost
And yes, it's fun to mess with people at this hour